The Colonial Origins of Ethnic Cleavages:
The Case of Linguistic Divisions in Zambia
Daniel N. Posner
Department of Political Science
University of California, Los Angeles
Los Angeles, CA 90095-1472
[email protected]
Prepared for LiCEP 3, Harvard University, 23-25 March 2001
Studies of the impact of colonialism on ethnic group formation are legion. Wolpe (1974),
for example, shows that the Igbo of Nigeria were a product of colonial-era boundary drawing.
Ranger (1989) explains how the Manyika of Zimbabwe were “created” by missionaries. Young
(1976) traces the origins of the Ngala of Congo to Henry Stanley’s misinformed labeling of the
people he encountered on the upper Congo river. And Gourevitch (1998) shows how the
distinction between Hutus and Tutsis in Rwanda was a product of Belgian administrative fiat.
Such studies, in addition to debunking primordialist assumptions about the origins of
ethnic loyalties, teach us much about how colonial policies and institutions shaped the salience of
particular identities and social divisions in the post-colonial era. What is missing these accounts,
however, is a story about how colonial administrative practices also affected the relative sizes of
groups and their spatial distribution around the state. What these studies lack, in short, is an
account of how colonialism was responsible not just for the emergence of particular ethnic groups
but for shaping the contemporary la ndscape of ethnic cleavages in the political system in which
they operate.
Why does the ethnic landscape matter? Because ethnic competition and conflict is about
not just who “we” and “they” are but about how politically powerful “we” are relative to “them.”
And relative political power is a function of relative size and geographic location. It is about
where a particular group fits within the larger landscape of other groups against which it is
competing. The conflict in Northern Ireland, for example, is about more than just the fact that
Catholics and Protestants feel strongly about who they are and want different things. It is about
the fact that one group comprises sixty percent of the population and the other comprises forty.
Without an understanding of the relative sizes of the groups, the nature and dynamics of the
conflict would remain opaque. Similarly, the origins of the civil war in Lebanon can be traced to
more than simply the fact that Christians, Sunnis and Shiites had different ideas about what the
Lebanese state should look like. Its origins lie, instead, in the fact that, by 1975, the relative sizes
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
of these three groups had changed dramatically since the power-sharing agreement that fixed their
relative power in 1943. It was the changing shape of the ethnic landscape, not the group
identities themselves, that was the cause of the conflict. Apart from relative size, a group’s
physical location within the country matters as well: the violence in Katanga in the 1960s cannot
be explained without accounting for how the Balunda, rather than local groups, came to dominate
the mining areas of the region.
If our goal is to gain insight into why individuals today feel attachments to particular
identity groups, then the usual accounts of the colonial origins of ethnicity probably go far
enough. But if our goal is to explain why the conflicts between ethnic groups in post-colonial
societies take the forms that they do, then we also need to explore how the policies of the colonial
state shaped the numbers, spatial distributions and relative sizes of the groups that are competing
in the ethnic landscapes of each country. In this paper, I provide an example of how this might be
done by tracing the origins of the contemporary map of language cleavages in Zambia.
Language is one of the two principal axes of social division in present-day Zambia
(Posner 1998). The fact that Zambia possesses four major language groups, rather than three or
twenty, and that each group is located in the parts of the country that it is, is critical for
understanding the dynamics of the country’s contemporary politics. The political dominance of
the Bemba-speaking community is directly related to its large size and the fact that it came to
dominate the politically crucia l mining towns of the Copperbelt. The comparative political
weakness of the Lozi, by contrast, stems from its smaller size and peripheral location away from
the industrial line of rail. A prerequisite for understanding Zambian politics, then, is to
understand why particular language groups came to have the sizes and geographical distributions
that they do. But the shape of this linguistic landscape would be inexplicable absent a careful
attention to the effects of decisions made during the colonial era in shaping it
In this paper, I show how specific actions and policies undertaken for purposes having
nothing to do with building identity groups or affecting their spatial distribution were responsible
2
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
for shaping the Zambian linguistic landscape today. I begin by explaining how the language map
was consolidated from one containing dozens of different groups to one containing just four.
Then I explain how these groups came to be physically located in the parts of the country that
they are.
The Consolidation of Languages in Zambia
At the time when the first Europeans reached the territory that comprises present-day
Zambia, language use corresponded almost perfectly with tribal affiliation. With the exception of
a handful of trading peoples that learned regional languages of commerce to facilitate their
trading efforts, Africans tended to speak the single language or dialect of their local community,
and each community had, more or less, its own language or dialect.1 At the beginning of the
colonial era, Northern Rhodesia was a Babel of languages.
By the end of the colonial era, patterns of language use had consolidated considerably.
As early as the late 1940s, Lord Hailey (1950) could report the emergence of a set of distinct
regional lingua francas in Northern Rhodesia.2 In the northern part of the protectorate, he noted,
“Chiwemba is practically the lingua franca of the tribes on the [central] plateau” (Hailey 1950,
part II: 138). In the east, the dialects of the various groups were “gradually being merged into
Chichewa, which is becoming the lingua franca of the [area]” (Ibid.: 131). In the west, Hailey
reported “an increasing fusion of language...Silozi is becoming the lingua franca of the [region]
and most of the tribes longest resident [there] understand it” (Ibid.: 92). Although Hailey was
silent on the extent of linguistic consolidation in the southern part of the protectorate, other
1 So close was the connection between community boundaries and patterns of language use that almost all
of the earliest tribal maps were, in fact, language maps, with linguistic distinctions serving as proxies for
more difficult-to-measure markers like cultural difference or traditional authority boundaries (Prins 1980:
33).
2 A comparison of Johnston's (1897) and Hailey's (1950) descriptions of the language situation in the
region gives a vivid indication of the degree of linguistic consolidation that took place during the half
century that separates the two accounts.
3
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
authors writing during this period noted the emergence of Citonga as a regional lingua franca in
that region (Colson 1962).
By the time of Zambian independence in 1964, Bemba (what Hailey called Chiwemba),3
Nyanja (Chichewa),4 Tonga (Citonga) and Lozi (Silozi) had achieved the status of first among
equals in so far as language use was concerned. By 1990, the first year for which reliable
information is available, fully 78.8 percent of the Zambian population used one of these four
languages as either their first or second languages of communication. 5 When we consider that
probably no more than a quarter of the population spoke these languages a century before, this
figure points to a remarkable – and remarkably rapid – consolidation of language use.6
The extent to which Bemba, Nyanja, Tonga and Lozi have emerged as dominant regional
lingua francas in Zambia is strikingly demonstrated in Figure 1, which compares the percentage
of the contemporary national population that uses one of the seventeen major Zambian languages
as their first or second languages of communication with the estimated percentage of the
population that used each language prior to the colonial era.7 As the chart indicates, the shares of
3 Technically, the prefix “chi” (or “ci” or “si”) is used before the name of the tribe to indicate the language
spoken by its members (e.g., Chibemba is the language of the Bemba tribe; Citonga is the language of the
Tonga tribe, etc.). Defering to common usage in Zambia, I drop the prefix when I refer to languages.
4 In Zambia, Chichewa and Chinyanja are different labels for the same language. There is no Nyanja tribe,
only a Nyanja language. Chewas (and also Ngonis) speak Nyanja – although very occasionally they will
refer to the language they speak as Chewa (or Ngoni).
5 The reports of the censuses of 1969 and 1980 provide tables with information on first languages of
communication only. The figures presented here were calculated by the author from the complete data set
of the 1990 census. To guard against double-counting, percentages of Zambians speaking Bemba, Nyanja,
Tonga or Lozi as a second language were calculated from the population of people who did not already use
one of these four languages as a first language of communication.
6 My estimate of the percentage of the population that used these four languages prior to the colonial era is
based on an assumption that, at that time, the speakers of each language were confined to the people who
were members of those tribes. The percentage of the population speaking Bemba, Nyanja, Tonga or Lozi
was therefore estimated from the percent of the Northern Rhodesian population belonging to each of these
tribes from the earliest date for which figures were available (1930). The reason I use 1930 figures rather
than 1990 figures for tribal proportions is because the population shares of some of the larger Zambian
tribes have increased significantly since the beginning of the colonial era. Using more recent figures would
have greatly overestimated the size of these tribes in the pre-colonial era, and thus the percentage of the
population that spoke their languages.
7 The figures reported for contemporary language use do not control for people who speak one of the
thirteen other languages as a first language in their counts of second language use. Estimates for language
use in the pre-colonial era were calculated as described in fn 5.
4
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
the population using Bemba, Nyanja, Tonga and Lozi in the pre-colonial period (indicated by the
white bars) were only slightly larger than the shares using other languages. By 1990, however,
the shares of the population that used these four languages (indicated by the black bars) far
exceeded the shares that used any of the other languages in the country. Close to 40 percent of
Zambians used Bemba as their first or second language of communication by that year, just over
30 percent used Nyanja, about 12 percent used Tonga and just under 10 percent used Lozi. After
these four languages, frequencies of language use dropped off considerably. The next most
frequently used languages, Tumbuka and Lamba, were used by only 3.8 and 3 percent of
Zambians, respectively.
Figure 1
Part of the reason that Bemba, Nyanja, Tonga and Lozi look as dominant as they do is
because, for reasons we will discuss, people came to le arn (and use) these languages in lieu of the
5
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
languages that were traditionally spoken by members of their tribes. Another important reason is
that, over time, Zambians came to develop language repertoires which included more than one
language of communication: usually one for home use (often their tribal language) and one or
more others for commercial or social exchanges with members of other tribes.8 Bemba and
Nyanja, and to a somewhat lesser degree Tonga and Lozi, emerged – along with English – as the
key languages that came to play this second role. Figure 1 suggests clearly that while a small part
of the growth of these four languages came from stealing shares from others, much of it came
from the acquisition of Bemba, Nyanja, Tonga or Lozi as second languages of communication. 9
Indeed, 25 percent of the people who speak one of these four languages do so as their second
language.10
How can this dramatic consolidation of language be explained? Some of it undoubtedly
took place between the end of the colonial era and 1990. Yet, to the extent that it did, it
represents the continuation of a trend whose origins lie in the period before independence.
Understanding how Zambia's linguistic map was transformed from one containing more than fifty
languages to one containing just four major ones requires that we delve into colonial history. As I
shall show, three colonial era forces in particular were responsible for the consolidation of
language use in Zambia: missionary activity, colonial education policies and labor migration.
Missionary Activity
Between 1885 and 1945, nearly two dozen different missionary societies set up shop in
Northern Rhodesia, establishing, between them, more than a hundred mission stations around the
8 For a discussion of the emergence of such language repertoires in Africa, see Laitin (1992).
9 Much of it also comes from the nearly total disappearance of the several dozen other languages that were
left out of Figure 1. Indeed, the fourteen languages included in the Figure only comprise 54 percent of total
language use in 1930.
10 This figure drops to 12.8 percent if we restrict our population of second language users to those who do
not already speak Bemba, Nyanja, Tonga or Lozi as their first language of communication.
6
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
country. 11 Although formally set up as evangelical outposts, these mission stations had as
important an impact on the territory's linguistic landscape as they did on its religious life.
Language was central to the missionary enterprise for a simple reason: in order to teach the
gospel, the Bible first had to be translated into the local language. And before the Bible could be
translated, the local language itself had to be written down. Early missionaries in Northern
Rhodesia therefore doubled as linguists. Many of them spent as much energy writing grammars,
compiling dictionaries and translating hymns, religious books and readers into new written
vernaculars as they did proselytizing (Johnston 1919; Doke 1945 and 1961).
Because the transcription of an African language required an enormous investment, “it
was only natural to amortize it by maximizing the diffusion of the standardized language forms to
neighboring groups, where possible” (Young 1976: 166). Missionary societies did this by
locating their stations, when they could,12 in areas where large numbers of people already spoke
the same language and by attempting, once they had completed preparing a grammar and
translating religious texts into a local language, to extend the use of that language to neighboring
peoples who spoke different, but structurally similar dialects.13 The local vernaculars into which
the Bible was first translated and for which grammars and dictionaries were first written were
thus “exported” from the domains where they were naturally spoken to adjacent areas, where they
gradually replaced or came to coexist with the languages that were previously in use. The fact
that the colonial administration encouraged missionary societies to concentrate their activities in
11 Figures are drawn from the “Northern Rhodesia Missionary Directory” provided as an appendix in Davis
(1967: 395-400), Snelson (1974) and Henkel (1985).
12 It was not always possible for missionary societies to choose the sites of their mission stations exactly as
they pleased. More often than not, they simply set up shop wherever they could get a toehold away from
stations already established on the ground or where they thought they would be received reasonably well by
the local chiefs.
13 Although economies of scale initially led missionaries to favor using the same language in many
settings, the logic of how missionaries in the field received funding from their home societies eventually
generated incentives for embarking on translations of new languages. In addition to justifying funding for
their work, translating new languages benefited individual missionaries by making them “indispensible” as
experts on particular peoples. The general trend of missionary-led language consolidation therefore was
offset to some degree by a counter-trend of (written) language proliferation. I thank Robert Rotberg for
pointing this out to me.
7
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
separate areas of the protectorate reinforced this “economy of scale” logic and facilitated the
process of regional language standardization.
The principal mechanism through which the exporting of vernaculars from one area to
another came to affect language use on the ground was native education. By 1925, the year that
the colonial government entered the education field for the first time, missionaries were operating
close to 2,000 schools throughout Northern Rhodesia with combined enrollments of more than
89,000 pupils.14 Even after that date, mission schools continued to play a central role in African
education. 15 As “essentially...literacy centers, supplemented by training in whatever skills or
interests the particular missionary possessed,” these early missionary schools had an enormous
effect on patterns of language use (Ragsdale 1986: 32). Over time, areas where mission stations
proliferated tended to coincide with increasing linguistic homogeneity.
Evidence supporting this link is presented in Table 1, which reports the results of a
statistical analysis of the relationship between missionary educational activities and the
homogenization of language use in Zambia’s 57 districts. The dependent variable is the ratio of
tribal to linguistic heterogeneity in each district.16 A perfect correspondence of tribal affiliation
and language use (as I argue was the case at the beginning of the colonial era) yields a value of 1.
Increasing linguistic homogenization is reflected in progressively larger and larger values greater
than 1, since the denominator (linguistic fractionalization) decreases as the numerator (tribal
fractionalization) remains constant. The ratio between the two thus serves as an excellent
14 Hall (1976: 83). Some caution is required in reading enrollment figures, as these numbers probably
reflect students formally registered in schools rather than the (considerably smaller) number that actually
attended school on a regular basis. Hall estimates that only two thirds of the enrolled students actually
attended.
15 According to enrollment figures in Snelson (1974: 296), it was not until 1940 that the number of African
students in mission schools was exceeded by the number in government-run schools. Even at that time,
mission schools had enrollments well in excess of 50,000 students.
16 Heterogeneity ratios were calculated from 1990 census data on tribal affiliation and language usage by
summing the squares of the percentages of every tribe (and language group) in the district that comprised
more than five percent of the district's population and then subtracting the sum from one.
8
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
indicator of the degree of linguistic homogenization that has taken place in the distric t since the
beginning of the colonial era.
Measuring the impact of missionary activities – the key independent variable – is
somewhat more complex. I began by identifying the present-day districts in which every mission
station established in Northern Rhodesia between 1880 and 1960 was located. 17 After recording
the number of stations in each district, I then weighted this value by the number of decades that
each station was in operation, thereby producing a district-level count of “station-decades.” Since
the mechanism through which missionary activities affected language use was via education, I
then weighted each station’s impact a second time by the educational commitment of the
missionary society with which it was affiliated. Societies strongly committed to African
education, like the Free Church of Scotland, the London Missionary Society or the Universities
Mission to Central Africa, received a score of 5. Societies with very weak commitments to
African education, like the Christian Missions in Many Lands or the South Africa General
Mission, received a score of 1. 18 These scores were then incorporated into the analysis to
produce an “educational commitment-weighted station decades” value for each district. To
smooth out the differences across districts, I then took the log of this value for use in the
regression analysis. Finally, because the logic of language homogenization in urban areas has a
logic of its own (to which we shall return), I included a dummy variable to control for the
urban/rural location of the district.
17 My key source for the presence and location of mission stations was Henkel (1985). When mission
stations were located on or near district boundaries, I drew a circle with a radius of 50kms around the
station and assigned “credit” to each district in proportion to the share of the circle located in each. My
thanks to Maria Dahlin for help in assembling this data set.
18 I am indebted to Robert Rotberg for his help in evaluating the educational commitment of the various
missionary societies.
9
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
Missionary Impact on Language Homogenization in Northern Rhodesia
Dependent Variable is Ratio of Tribal and
Linguistic Heterogeneity in the District
1.62**
(0.527)
Urban
Log of Educational
Commitment-Weighted
Station Decades
0.758*
(0.305)
Constant
0.551
(0.435)
R2
0.15
* significant at the 0.05 level
**significant at the 0.01 level
N=57
Table 1
The results of the analysis suggest, first, that urban location matters. For reasons we shall
explore, urban settings generate a strong standardization of language use. Our more important
finding, however, is that, controlling for urban/rural location, districts that were home to
missionary societies committed to the education of Africans had significantly higher ratios of
tribal to linguistic heterogeneity in 1990 (i.e., greater evidence of language homogenization) than
those that were not. The evidence thus confirms the argument: missionary activity led to the
consolidation of language use.
Colonial Education Policies
The homogenization of language use that was begun by the missions was reinforced and
expanded by the policies that the colonial government adopted when it took over primary
responsibility for African education after 1925. The most important such policy was the decision
taken by the Advisory Board on Native Education in July 1927 to simplify the administration’s
10
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
job by adopting just four languages – Bemba, Nyanja, Tonga and Lozi – as languages of
instruction in native schools.19 While the Board recognized that some pupils would initially have
to continue to receive their early primary education in languages other than these four, it was
assumed that eventually Bemba, Nyanja, Tonga and Lozi would become, with English, the sole
languages of instruction at all levels.20
To give effect to this new policy, an African Literature Committee – the first of its kind
in Africa – was established to promote the publication of secular school books in each of these
four languages. Between 1937 and 1959, the Committee and its successor, the Joint Publications
Bureau of Northern Rhodesia and Nyasaland, published or reprinted 484 titles, the vast majority
of which were in Bemba, Nyanja, Tonga, Lozi or English. 21 The dominance of these languages
among the books produced by the Bureau is readily apparent in Figure 3, which shows the
number of volumes published in each major Zambian language in the years 1949-1959. In all,
more than 1.3 million volumes were printed in Bemba, Nyanja, Tonga or Lozi in these years,
compared with just 127,000 in all other local African languages combined. 22
19 Extract from Minutes of the Ninth Meeting of the Advisory Board on Native Education,” July 1927
(Zambia National Archives, file RC/1680).
20 In choosing these four languages, the government estimated that they could be “used in the earliest
stages of education for 55 percent of the natives of Northern Rhodesia. For another 25 percent, books in
these languages could probably be introduced without difficulty from Standard I [grade two] onwards.
These and the other 20 percent will have to continue to rely on primers and translations of the Scriptures
produced by local missionaries for the first years of schooling.” Northern Rhodesia Annual Report Upon
Native Education, 1927, quoted in Ohannessian and Kashoki (1978: 287).
21 Publications Bureau of Northern Rhodesia and Nyasaland, Annual Report for 1959. Although these
figures also include a small number of books published for the Nyasaland market, the majority of these
books were published in Nyanja and were therefore also useable in Northern Rhodesia. In addition to
language texts, the books published by the Bureau covered a variety of subjects, from tribal history and
African folk lore to child care and village sanitation.
22 Evidence that books in Bemba, Nyanja, Tonga and Lozi were demanded as well as supplied comes from
“best seller” lists provided in the Publications Bureau's annual reports. These lists reveal that every single
one of the best-selling books in these years was written in one of these four languages or English
(Publications Bureau of Northern Rhodesia and Nyasaland, Annual Reports for 1949-58). No “best-seller”
list is given in the 1959 report.
11
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
Figure 3
The demand for vernacular language books was fueled by the growth of African
education in the 1940s and 1950s. As Figure 4 demonstrates, both school enrollments and
government expenditure on African education exploded at the end of the 1930s. Whereas only 25
percent of school age children were estimated to have been attending school in 1924, the share
attending school in 1945 reached 75 percent in some rural districts and exceeded that number in
many of the major towns.23 Major efforts were also made during this period to promote literacy
among adults, particularly on the Copperbelt.24 Although many areas of the country remained
educational backwaters, the general trend was for more and more Africans to be exposed to
23 The 1924 estimated is from the report of the Phelps-Stokes Commission, cited in Gadsden (1992); 1945
figures are from Snelson (1974: 240). High rates of school attendance in the towns were an outcome of the
government's decision, after the 1935 Copperbelt riots, to make education compulsory for children between
12 and 16 years old in the mining towns. Doing so, it was felt, would “keep them out of trouble” (Greig
1985: 42).
24 By the late 1950s, more than 4,500 adult men and women attended educational classes each day on the
Copperbelt (E. C. Bromwich, “General History of Roan,” 1963, ZCCM Archives). See also Hay (1947).
12
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
Figure 4
formal education and for the quality of that education to improve gradually over time. And the
fact that the medium of instruction in these schools and literacy courses was Bemba, Nyanja,
Tonga or Lozi meant that every newly literate student that did not already speak one of these
languages as their mother tongue became a convert to one of these language groups.
The impact of formal education on language consolidation was reinforced by trends in
the popular media. In the 1936, the colonial government began publishing the African newspaper
Mutende as a response to the Watch Tower Movement, whose authority-questioning literature
was, by that time, in wide circulation around the protectorate. Published in Bemba, Nyanja,
Tonga, Lozi and English, Mutende reached a peak circulation of 18,000 during the war years
(Hall 1976: 76, fn 11). As with most African newspapers, however, the number of people that
13
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
Mutende reached was many times its circulation. During the 1950s, the mining companies also
began publishing monthly magazines for their workers. Although most of the stories in these
magazines were initially in English, increasing numbers of articles appeared over the years in
Bemba and, to a lesser degree, Nyanja.
Even more important than newspapers in bolstering the dominance of Bemba, Nyanja,
Tonga and Lozi in Northern Rhodesia was radio broadcasting. Thanks to the invention and rapid
proliferation of the “Saucepan Special,” an inexpensive battery-operated radio set developed
specifically for the Northern Rhodesian African population by the protectorate's Director of
Information, Harry Franklin, many thousands of Africans had access to radio in Northern
Rhodesia by the 1950s (Franklin 1950). Largely because it knew it had such a big audience of
African listeners, the Northern Rhodesian Broadcasting Service was the first radio service in
Africa to allocate significant air time – fully 72 percent in 1952 – to programming in vernacular
languages (Mytton 1978: 209). Bemba, Nyanja, Tonga and Lozi were chosen, with English, as
the languages of Northern Rhodesian broadcasting. Because radio reached such a large
population, the choice of these languages had a critical impact on patterns of language
consolidation in the country – more, even, than the education system, which directly touched
fewer people. “Over time,” Spitulnik notes, “the selection and dominance of [these] four
languages became mutually reinforcing” (1992: 340). Africans learned Bemba, Nyanja, Tonga
and Lozi by listening to the radio and, having learned them, demanded more programming in
these languages.
Labor Migration
The third major force that contributed to the consolidation of language use in colonial
Northern Rhodesia was labor migration. From as early as the turn of the century, the mines and
farms of Southern and Central Africa demanded large numbers of able -bodied African laborers.
In its capacity as the administrator of Northern Rhodesia, the British South Africa Company
14
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
(BSA Co.) controlled one of the most important territories from which these laborers were
sought.25 In its capacity as the owner of both the Southern Rhodesian mines and the railroad
system that served the Katanga ore body, the Company also controlled some of the key
enterprises that stood to suffer if the demand for African labor was not met.26 In the interest of
maximizing its income, then, the BSA Co. took advantage of its administrative powers in
Northern Rhodesia to put policies in place that would ensure that an abundant supply of African
laborers would be available for the region's industries.
The principal instrument used by the Company for this purpose was native taxation. By
“consciously [setting] the rate of tax at a level that would successfully draw African males away
from their homes to the usually distant centres of white agriculture and industry,” the
Administration forced thousands of Northern Rhodesians out of their villages (Rotberg 1965: 41).
“In theory,” Gann (1958: 84) writes,
the tax could be earned by working only a fortnight in Southern or a month in
Northern Rhodesia for wages. These calculations, however, took no account of
traveling time nor of the fact that the migrant could not save all his pay. Besides,
it was impossible for every male to undertake an annual journey, for there was
much work women could not do. Thus every tribesman had to bring back
enough to pay for some of his fellows; and lengthy journeys away from the
village...became essential in order to meet their heavy obligations.
As early as the second decade of the century, the flow of migrants from rural Northern Rhodesia
to the mines and farms of Southern Rhodesia, South Africa and Katanga was so great that native
commissioners in Northern Rhodesia began to complain that their districts were becoming
“denuded of their menfolk” (Ibid.: 86). By 1938, absentee rates of working age males equaled or
25 Just prior to the turn of the century, the Chief Native Commissioner for Matabeleland (Southern
Rhodesia) wrote that “there is not sufficient labour in Matabeleland to supply the various mines when they
are at work; we must look to the outside provinces for our supply...Our best source of supply is from the
north of the Zambezi.” Reports on the Administration of Rhodesia, 1897-98, quoted in Meebelo (1971:
81).
26 In addition to its business concerns, the Company also had an economic interest in making sure that the
labor requirements of the European settler populations of the Northern and Southern Rhodesian territories
were met. Particularly in years when the Company's mining and railroad properties yielded weak profits, a
significant portion of the BSA Co.'s revenues came from the sale of land to White farmers and from the
royalties received on small-scale mining ventures undertaken by European settlers (Gann 1958: 79).
15
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
exceeded fifty percent in seven of Northern Rhodesia's thirty-three rural districts and approached
that level in six others.27 Roberts (1976: 191) estimates that, by approximately that date, “more
than half the able -bodied male population of Northern Rhodesia was working for wages away
from home.”28
For the first two and a half decades of the century, the vast majority of Northern
Rhodesian labor migrants traveled outside the territory to find work. But by the end of the 1920s,
Northern Rhodesia's own copper industry also began to demand large numbers of African
laborers. At the urging of the local mining companies, the colonial administration, which had
taken over control of Northern Rhodesia from the BSA Co. in 1924, began taking an increasing
interest in making sure that its own industries would be adequately supplied with labor. To make
certain that they would be, the government put in place a series of measures to channel Northern
Rhodesian manpower to its own mines. These intra-territorial labor flows that resulted were to
have a profound effect on the shape of the country’s language map.
A key effect of the administration’s policies was to bring an ever-increasing number of
migrants from the hinterland to the industrial line of rail. Once there, their patterns of language
use changed. Since both productivity on the job and everyday interactions in the social sphere
required that people be able to communicate with each other, a single language naturally emerged
as a common medium of communication in each urban area. And once such an urban lingua
franca was established, new migrants who spoke other languages had incentives to learn the
common language in order to participate in the activities of urban life. A strong tendency
towards linguistic homogenization therefore emerged along Northern Rhodesia’s line of rail.
27 Report of the Commission Appointed to Enquire into the Financial and Economic Position of Northern
Rhodesia, 1938, cited in Hellen (1968: 96).
28 Roberts' estimate is supported by figures from the 1935 Northern Rhodesia Report on Native Affairs.
Drawing on these figures, C. F. Spearpoint, the compound manager at the Roan Antelope Copper Mine
(RACM), calculated that 40.1 percent of all taxable males in Northern Rhodesia were employed away from
their villages: 6.4 percent on the Northern Rhodesian mines, 17 percent outside of Northern Rhodesia, and
16.7 percent in non-mining work within Northern Rhodesia (memorandum from Spearpoint to RACM
16
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
And the policies that brought thousands of laborers there thus contributed significantly to the
country-wide consolidation of language use.
We already saw evidence for this trend of urban linguistic homogenization in the large
and highly significant coefficient on the “urban” variable in Table 1. Additional evidence comes
from 1990 census figures, which reveal a wide gulf between the ratios of tribal and linguistic
heterogeneity in urban and rural districts of the country. Whereas Zambia's ten urban districts
have an average ratio of 2.23 (indicating that they are much more heterogeneous tribally than
linguistically), its forty-seven rural districts have an average ratio of 1.5 (indicating that they are
also more heterogeneous tribally than linguistically, but far less so than the urban districts).29
Since almost all urban residents during the colonial era started out as rural dwellers, the different
ratios suggest that the act of moving from a rural to an urban environment affected the likelihood
that a person would speak one of the country's principal lingua francas. The data thus confirms
that the experience of migration contributed to the standardization of language use.
The Shape of the Linguistic Map in Zambia
Missionary activities and colonial education policies help to explain how the dozens of
African languages spoken in the pre-colonial period gave way to four principal languages of
communication by the time of Independence in 1964. Labor migration helps to explain how this
trend of language consolidation was carried over from the rural to the urban areas. But the causal
mechanisms we have discussed thus far provide few clues as to why the populations that speak
each of these languages came to be physically located in the areas of the country that they are. To
be sure, the decision of the colonial administration in 1927 to adopt Bemba as the language of
General Manager F. A. Ayer, 16 November 1936, ZCCM Archives, file WMA 139). If we figure that only
80 percent of all taxable men are actually “able-bodied,” then we arrive at almost exactly Roberts' estimate.
29 Although some of the linguistic homogenization that these figures reflect undoubtedly took place after
the colonial era, the data is still suggestive of the effect of labor migration on language consolidation. Here
the important factor is the difference between the urban and rural averages, not when the homogenization
took place.
17
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
instruction in the northern part of the territory, Nyanja in the east, Tonga in the south and Lozi in
the west does tell us something about the spatial distribution of these language communities. But
it does not explain why different parts of the industrial line of rail came to be dominated by the
groups that they were. Nor do the variables we have described provide insight into why the
political coalition that each language group comprises came to have the size – and thus political
clout – that it does. To understand these aspects of the shape of Zambia’s contemporary
linguistic map we need to look not just at the fact of labor migration but at its pattern.
If Figure 1 told a story about the consolidation of language use, Figure 2 suggests a story
about the distribution of language groups around the country. The map identifies districts in
which, according to 1990 census data, more than 80 percent (shaded in dark colors) and 40
percent (shaded in lighter colors) of the population spoke Bemba, Nyanja, Tonga or Lozi as their
first or second languages of communication in 1990. As the map makes clear, each language
predominates in a particular region of the country: Bemba in the north, Nyanja in the east, Tonga
in the south and Lozi in the west.30
30 Of course, part of the reason that each of these language groups predominates in the region that it does is
because this is where the tribe that originally spoke that language was located. But the distribution of that
tribal language to its surrounding rural hinterland and urban satellite requires explanation.
18
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
The map also makes clear that, while Zambia's urban areas may share similarly
homogeneous patterns of language use (hence the dark colors in the towns along the rail line), the
particular languages that came to dominate each urban setting vary. Bemba emerged as the urban
lingua franca on the Copperbelt and in the mining town of Kabwe.31 Nyanja serves this purpose
in Lusaka.32 And Lozi became the lingua franca of Livingstone.33 Thus, while the
circumstances of urban life may have guaranteed the emergence of an urban lingua franca in
every town, they did not guarantee that the same language would serve this purpose in each one.
31 According to 1990 census figures, 90.2 percent of Copperbelt residents (excluding Ndola Rural district)
spoke Bemba as their first or second language of communication. In Kabwe, the figure is 77.1 percent.
32 In 1990, 78 percent of Lusaka residents spoke Nyanja as their first or second language of
communication.
33 Lozi's position in this regard has been undermined since independence by Tonga-speakers who have
aggressively pushed the use of Tonga rather than Lozi on the grounds that, as the capital of a province
whose people are overwhelmingly Tonga-speaking, Livingstone should rightly be a Tonga-speaking town.
19
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
The language that emerged as dominant in each urban area was determined by patterns of
migration: the region of the country that supplied the largest percentage of the urban area's
migrant population also supplied the language that the urban population came to speak. To
explain why specific towns came to be dominated by particula r languages, then, requires that we
account for patterns of migration from particular rural hinterlands to particular urban population
centers. Doing so requires that we turn again to the policies of the colonial administration and the
mining companies.
Linking Towns and Languages
The link between government and mining company policies and the emergence of
specific languages as lingua francas in particular urban areas is most evident in the case of the
Copperbelt and Kabwe (Broken Hill), where conscious policies were put in place to encourage
migration from Bemba-speaking rural areas to the Northern Rhodesian mines. In other parts of
the protectorate, the government's concern was simply to stimulate migrant labor flows of
sufficient magnitude to allow taxes to be paid in the rural areas. But in the Bemba-speaking
northeast,34 the government and the mining companies conspired not simply to encourage
outward labor migration but to make certain that this migration would be channeled to the
domestic copper mines. The dominance of the Bemba language in the mining towns was a direct
outcome of these policies.
The government and mining companies' particular interest in channeling laborers from
the Bemba-speaking northeast to Kabwe and the Copperbelt stemmed from their desire to ensure
that African labor on the Northern Rhodesian copper mines would be both plentiful and cheap.
Several factors conspired to ensure that outflows of work-seeking migrants from the Bembaspeaking heartland would satisfy this first condition. A combination of very poor soils, the
By 1990, only 44.4 percent of Livingstone residents said that they spoke Lozi as their first or second
languages of communication.
20
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
presence of the tsetse fly in much of the area and the great distance that agricultural products had
to travel to the markets of the line of rail ruled out cash cropping or animal husbandry as a means
of earning money to pay one's taxes. The dearth of European settlers in the area also meant that
local employment opportunities were limited. Thus, to an even greater degree than in most other
regions of the protectorate, residents of the northeastern part of Northern Rhodesia had few
alternatives to labor migration (Richards 1939; Moore and Vaughan 1994).
Geography also affected – and allowed the government and mining companies to
manipulate – the price of labor from Bemba-speaking areas. Of all the regions of Northern
Rhodesia, the Bemba-speaking northeast was located furthest from the South African and
Southern Rhodesian mines and closest to the alternative mining center of Katanga. This meant
that established labor migration routes from Bembaland ran from east to west (to Katanga) rather
than from north to south (to Southern Rhodesia and South Africa), as they did in the other parts of
the protectorate. At the time that the Northern Rhodesian mines began production in the late
1920s, the Bemba-speaking northeast therefore constituted a labor reserve where the Copperbelt
companies faced relatively little competition from their better-paying South African and Southern
Rhodesian rivals.35 By focusing the bulk of their labor recruiting efforts on the northeast, and by
protecting this region from competition by labor recruiters from other Southern African mining
centers, the Northern Rhodesian mining companies could – and did – keep the wages they paid at
less than half the prevailing rates in the rest of the region.36
34 By the “Bemba-speaking northeast” I refer roughly to the present-day Northern and Luapula provinces.
35 The cessation of large-scale labor recruitment from Northern Rhodesia by the Union Miniere du Haut
Katanga in the early 1930s removed the other potential rival for African labor from the northeast and freed
up large numbers of Bemba-speaking men who might otherwise have followed traditional migration routes
to Katanga. Because the Katanga mines paid very low wages – significantly lower than those paid by the
Copperbelt companies, according to data presented in Perrings (1979: 258) – the Northern Rhodesian
companies did not have to worry about matching the wages to which former Katanga workers were
accustomed.
36 In 1933, for example, wages on the Rand mines averaged 3/- per shift, while wages in the Northern
Rhodesian mines averaged just 1/- per shift (Report of the [Northern Rhodesia Native Labour Association]
Manager's Visit to Livingstone in Reference to Labour Matters, 20 September 1933, ZCCM Archives, file
WMA 139).
21
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
Evidence from internal mining company sources confirms that wage calculations were at
the center of the strategy to recruit African labor from the northeastern part of the protectorate.
When the Union of South Africa withdrew its voluntary embargo on recruiting African labor
from Northern Rhodesia in 1933, the Copperbelt mining companies became alarmed that the
already steady flow of Northern Rhodesian migrants to the south might increase to the point
where the protectorate's own labor supply would become insufficient to meet local needs. If this
were to happen, the manager of the Northern Rhodesian Native Labour Association (NRNLA),
the organization responsible for recruiting labor for the country’s mines, worried, “our only
means of meeting competition” from the South African mines would be to “increas[e] our wage
scale...and the resultant extra costs to our mines would be £225,000 on a year...This extra cost
would obviously result in a reduction of profits.”37 Anxious to avoid a situation of competitive
recruiting with the Rand mines that might bid up African wages in the Northern Rhodesian
mining areas, the Copperbelt companies told the Northern Rhodesian government that they would
be “prepared to abandon all recruiting activities in the [western part of the protectorate] and to
leave that area open to the Rand recruiters provided that, in turn, they would not encroach on any
other Northern Rhodesian districts.”38 The goal, the NRNLA manager made clear, was to make
certain that the introduction of South African labor recruiting would “not materially diminish our
labour resources in the Northern and Eastern areas.”39 Recognizing that their own revenues
depended on the profitability of the Copperbelt mines, the government agreed to protect these
areas by adopting the NRNLA's proposal to limit Rand recruiters to the western districts.
The mining companies efforts to forge links between the Copperbelt and Kabwe and the
Bemba-speaking northeast during the 1930s went far beyond the protection of the northeast from
foreign labor recruiting agents. When the NRNLA began recruiting labor for the Copperbelt and
Kabwe-based mines in 1930, the largest share of recruits were intentionally drawn from Bemba37 Ibid.
38 Ibid.
22
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
speaking areas.40 When the mines came back on line in the mid-1930s after the depression, the
first rural district officers that were advised to lift their restrictions on issuing passes for migrants
to travel to the Copperbelt were those in Kasama, Fort Rosebery and Abercorn, three of the most
populous Bemba-speaking districts.41 And when, at the insistence of the colonial government,
the mining companies agreed in 1939 to underwrite the construction of rest camps along labor
migration routes to the Copperbelt and Kabwe, nine of the ten that were constructed were built
along routes from the Bemba-speaking northeast.42 This decision represented both a recognition
of the nature of existing migrant flows and an investment in perpetuating them.
All of these efforts led, by the end of the 1930s, to the establishment of entrenched labor
migrancy routes between Bembaland and the Northern Rhodesian mining centers. In 1937, fully
51 percent of the African workers employed at the Copperbelt's three largest mines were from the
Bemba-speaking northeast.43 This is a remarkable figure when we consider that the northeast
contained only between 20 and 25 percent of the protectorate's total population. By 1961, if we
exclude alien laborers from the count, the proportion of Copperbelt laborers from this area was
over 60 percent.44 This number was more than sufficient to tilt the linguistic balance in favor of
Bemba as the urban lingua franca of the mining towns. Given that, by the time of independence,
the mining areas contained nearly a quarter of Zambia's total population, the role of the
government and the mining companies in establishing Bemba as the lingua franca of the mining
towns had a profound effect on the shape of the contemporary Zambian linguistic – and also
political – map.
39 Ibid.
40 In part, this was an outcome of stereotype-driven assumptions that the martial history of the Bemba tribe
made Bembas good laborers. On the effects of such stereotypes on labor recruiting policies and Copperbelt
social relations, see Luchembe (1992) and Siegel (1989).
41 Letter from NRNLA Manager A. Stephenson to RACM Manager F. Ayer, 4 August 1933 (ZCCM
Archives, file WMA 139).
42 ZCCM Archives, file WMA 135.
43 Pim Report, cited in Berger (1974: 15).
44 Chamber of Mines Year Book, cited in Harries-Jones (1965: 130). If alien laborers are included, the
percentage of migrants from the Bemba-speaking hinterland drops to just over 46 percent.
23
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
The government's role in shaping the urban lingua francas that developed in Livingstone
and Lusaka, the territory's two other major urban population centers, was less direct than it was in
Kabwe or the Copperbelt. Although the colonial administration did have a hand in encouraging
the Lozi- and Nyanja-speaking migrants that came to dominate these areas to leave their rural
homes in search of wage employment – it did this everywhere in the hopes of enabling rural
Africans to meet their tax obligations – the administration did little specifically to encourage
these migrants to settle in Livingstone or Lusaka. The fact that large numbers of Lozi- and
Nyanja-speaking migrants eventually settled in (and lent their languages to) these towns was,
more than anything else, an artifact of the limited transportation infrastructure that was available
at the time to take these migrants to the labor centers to which the colonial government
encouraged them to travel.
Livingstone became a Lozi-speaking town because it served as the railroad terminus for
trains heading south to the mines of South Africa and Southern Rhodesia and because, from
before the turn of the century, the Lozi-speaking hinterland had served, with the administration's
encouragement, as one of the major Northern Rhodesian labor reserves for these southern
mines.45 The construction of the Mulobezi-Livingstone railway, which covered half the distance
from the center of Barotseland to the south-bound rail terminus at Livingstone, further
contributed to the close link between Lozi migrants and the town by greatly reducing the cost of
and time required for migration to the Livingstone railhead. 46 Thus, when Livingstone-based
industries like the Zambezi Sawmills began to require larger numbers of workers in the 1930s and
1940s, existing flows of south-bound migrant workers through Livingstone guaranteed that the
most plentiful supply of African laborers would be from Lozi-speaking areas.47 By 1956, two-
45 The link between the Lozi-speaking western portion of the protectorate and the Rand mines was
cemented after 1940 when the Northern Rhodesian government began permitting the Witwatersrand Native
Labour Association (WENELA) to recruit laborers from this area for the South African gold fields.
46 Before the construction of the railway, migrants had to travel the distance on foot (Philpott 1945).
47 In addition, at least some Lozis were recruited directly to Livingstone by the Northern Rhodesian
government. Intensive missionary education work undertaken in Barotseland at the behest of the Lozi
24
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
fifths of Livingstone's population was from Barotseland (McCulloch 1956). Although Lozispeakers were never a majority, they were by far the largest group in the town, and the Lozi
language became Livingstone's urban lingua franca.
Nyanja became the lingua franca of Lusaka for similar infrastructure-related reasons.
Before the opening of the Northern Rhodesian copper mines, the colonial administration
encouraged Nyanja -speaking migrants from the eastern part of Northern Rhodesia to forge links
with the mines and farms of the south, particularly those in Southern Rhodesia. After the late
1920s, the government and mining companies also began recruiting men from this area to the
Copperbelt. The nearly total absence of north-south roads directly linking eastern Northern
Rhodesia with either Southern Rhodesia or the Copperbelt, however, meant that Nyanja -speaking
migrants from the east had to travel along the Great East Road until it met the rail line in Lusaka
before they could turn south or north towards their ultimate destinations.48 Until the late 1940s,
Lusaka served as little more than a way-station for these Nyanja -speaking migrants from the east.
By the 1950s, however, both the colonial administration, which had moved to Lusaka in 1935,
and the various businesses and industries that had located there began to demand significant
numbers of African laborers. Since Nyanja -speakers from along the Great East Road comprised
the majority of the available work force, they became the majority of settlers in the town. By
1959, they made up more than 40 percent of Lusaka's population (Bettison 1959). And because
Nyanja-speakers made up the largest single language group, their language became the lingua
franca of Lusaka.
In the case of all three of these urban areas, early patterns of labor migration – generated
in one case by conscious policy and in the two others by infrastructure constraints – led to the
entrenchment of migration links between each town or segment of the line of rail and a different
paramount chief meant that Lozis were among the best educated Africans during the first several decades
of the century, and this made them especially sought after for the colonial civil service. During the brief
period (1924-1935) that the colonial administration was situated in Livingstone, well-educated Lozis were
recruited to work as government clerks and interpreters.
25
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
rural hinterland. In each case, the language of the rural hinterland became the lingua franca of
the urban area. These linkages are clearly evident in Figure 2, in which major roads and rail lines
clearly connect the blue-shaded the Bemba-speaking northeast with the blue-shaded Kabwe and
Copperbelt, the red-shaded Nyanja -speaking east with the red-shaded Lusaka and the greenshaded the Lozi-speaking west with the-green shaded Livingstone.
Only the Tonga-speaking language area, which straddles the southern half of the line of
rail, does not have its own urban enclave. In fact, the single major town located within the Tonga
zone, Livingstone, is a Lozi-speaking rather than Tonga-speaking area. The reason for the
absence of a Tonga-speaking urban satellite is that, in contrast to people living in the Bemba-,
Nyanja- or Lozi-speaking rural areas, Tonga-speakers had abundant opportunities for local
employment, either on the many European farms located along the line of rail or as individual
cash croppers or cattle herders. The fact that Tongaland was bisected by the railway line meant
that any crops or cattle that were raised there could be easily (and inexpensively) transported to
markets on the Copperbelt or in Southern Rhodesia. Tax obligations could therefore be met by
Tongas by hiring themselves out to locally situated European farmers or by engaging in rural
agricultural production. Migration to distant urban employment centers was unnecessary and, for
the most part, avoided. 49
Figure 2 also reveals an additional map-shaping effect of urban migration. In addition to
affecting patterns of language use in the towns, the linking of each urban area with a specific rural
hinterland also affected patterns of language use along the routes that the migrants traveled.
Notice that the Bemba- and Nyanja -speaking zones extend like fingers from their linguistic
48 The road was constructed in 1928 (Henkel 1985: 10).
49 According to data presented in Hellen (1968: 99), fully 62 percent of the taxable males from the Tonga-
speaking Southern Province were at work locally in 1961. Comparable figures for the other major rural
areas were just 20 percent for the combined Bemba-speaking populations of Northern and Luapula
provinces, 21 percent for the Nyanja-speaking Eastern Province and 25 percent for the Lozi-speaking
Barotse Province.
26
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
epicenters towards their respective urban satellites.50 Over time, the languages spoken in the
rural hinterlands were diffused to the areas located between the migrants' homes and their urban
destinations.
Northwestern Province: The Exception that Proves the Rule
A final, quite obvious, point to note about the map provided in Figure 2 is that the entire
northwestern portion of the country is unshaded, signifying that none of the four major lingua
francas took hold in this area. Not only have Bemba, Nyanja, Tonga and Lozi been unable to
penetrate the northwest, but no single local language has managed to emerge as a dominant
regional lingua franca for the region. Although Lunda, Kaonde and Luvale enjoy first-amongequals status in the area, none of them has approached the positions of dominance that Bemba,
Nyanja, Tonga or Lozi have established in their respective rural regions.51 From the standpoint
of language use, northwestern Zambia constitutes the great exception to the trend of policy-driven
linguistic standardization in colonial Northern Rhodesia. As we shall see, it also constitutes the
exception that proves the rule. For it is precisely the absence of the factors that explain the
linguistic consolidation that took place in the rest of the country that explains the preservation of
multiple languages of communication in Northwestern Province.
First, missionary societies, which played such a central role in consolidating patterns of
language use in Northern Rhodesia, could not play this role in the northwestern part of the
protectorate for the simple reason that very few missions were located there. Moreover, until
50 The absence of a similar bulge in the Lozi-speaking rural penumbra towards Livingstone is largely an
artifact of the highly aggregated district-level data from which the map in Figure 2 is built. A more finegrained map would reflect the influence of Lozi-speaking migrants on patterns of language use in the area
between Livingstone and the Lozi epicenter.
51 According to 1990 census figures, Lunda, Kaonde and Luvale are spoken as a first or second language
by 34.3, 28.1 and 23.9 percent of the population of Northwestern Province, respectively. By comparison,
Bemba is spoken as a first or second language by 70 percent of the populations of Northern and Luapula
Provinces, Nyanja is spoken as a first or second language by 86 percent of the population of Eastern
Province and Tonga is spoken as a first or second language by 79 percent of the population of Southern
Province.
27
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
1931, the few missionary societies that did build stations in the northwest tended to be
evangelical groups that were far less concerned with African education and language work than
their colleagues from the main-line denominations that set up shop in other parts of the territory
(Gadsden 1992: 104).52 Indeed, while the average “educational commitment-weighted station
decades” value for all rural Zambian districts is 33.5, the average for the six districts located in
Northwestern Province is just 8.2 – significantly lower than any other rural province.
Second, colonial education policies, which had bolstered the positions of Bemba, Nyanja,
Tonga and Lozi by designating them as vernaculars of instruction in schools and making them
priority languages for book publishing, made no special arrangements for the major northwestern
languages and thus did nothing to stimulate their diffusion outside their tribally-defined core
areas. As Figure 3 revealed, only a tiny fraction of the books published by the government in the
1950s were in Lunda, Kaonde or Luvale. Although radio broadcasts in Lunda and Luvale were
begun in 1954,53 these two languages were only allocated half the air time of Tonga and Lozi and
less than a third of the air time of Bemba and Nyanja (Mytton 1978: 210). The inclusion of these
languages on the broadcasting roster, while no doubt contributing to their prestige (Spitulnik
1992: 341-42), thus probably had little effect on language standardization in the northwest.
Finally, although large numbers of men from northwestern Northern Rhodesia migrated
to the Copperbelt – along with the Bemba-speaking northeast, the northwest had the largest
percentage of males at work outside their villages of any region of the protectorate 54 –
Northwesterners never came close to equaling the number of Bemba-speaking migrants in even
the most northwestern of the mining towns.55 Lunda, Kaonde or Luvale therefore never
52 According to Doke (1945), the first grammars or dictionaries in Lunda, Kaonde or Luvale did not appear
until the 1920s. Bemba grammars and language handbooks, by contrast, dated from 1904, Nyanja sources
from 1885, Tonga primers from 1906, and the first Lozi (Kololo) grammar from 1914.
53 Kaonde was added as a seventh language at independence in 1964.
54 Figures, for 1961, are from Hellen (1968: 99).
55 According to figures from the Chamber of Mines Year Book for 1961 cited in Harries-Jones (1965: 130),
migrants form the northwest constituted only 8 percent of the non-alien workforce on the Copperbelt
mines. Of course, even if they had constituted a larger percentage, the variety of languages spoken by
28
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
challenged Bemba as the Copperbelt's lingua franca. Rather than recruit other language speakers
to their own vernacular (as Bemba-speakers did in the mining towns, Lozi-speakers did in
Livingstone and Nyanja -speakers did in Lusaka), migrants from the northwest either adopted
Bemba as a replacement for their own mother tongue or expanded their linguistic repertoire to
include Bemba as a supplementary language. Labor migration therefore did not have the same
effect on the diffusion of the northwestern languages that it did for Bemba, Nyanja or Lozi.
Conclusion
Group distributions are often taken as givens, just as the existence of the ethnic groups
themselves once were. But for the same reasons that groups cannot be seen this way, neither can
the contours of the ethnic landscape in which they are located. And just as the as the existence of
ethnic groups like the Igbo, Manyika, Ngala and Hutu can be traced to the policies of colonial
regimes, so too, in many cases, can the physical locations of groups and their sizes vis-à-vis other
actors in the political system. In this paper, I have illustrated this point by showing how the
contours of Zambia’s contemporary language map – the number of groups it contains, their
relative sizes and their spatial distributions around the country – can be explained by specific
policies and actions taken by missionary societies, the colonial administration and the Northern
Rhodesian mining companies during the colonial era.
Apart from their complementary effects on the nature of the Zambian linguistic
landscape, one of the noteworthy characteristics shared by all of these policies and actions is that
none of them were motivated by a desire to shape that landscape, or even a recognition that this
might be one of their effects. They were motivated, instead, by concerns about saving costs
and/or facilitating administration. The missionaries needed to translate the Bible. But, because
migrants from the northwest would have presented a significant obstacle to a northwestern linguistic
takeover of the Copperbelt.
29
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
translation was costly, they sought to minimize the cost of their efforts as much as possible. They
therefore sought to extend the boundaries of the language communities in which they were
working beyond their original confines to amortize their costs over a larger population. The
colonial government’s actions can be seen in similar light. The administration knew it needed to
become involved in African education but sought to simplify its task by reducing the number of
languages it would use in textbooks and classroom instruction. It therefore selected and
promoted the expansion of just four. The government also conspired with the Northern
Rhodesian mining companies to ensure the profitability of the local copper industry. They did
this by implementing tax policies that would encourage the steady flow of labor migrants to the
urban mining centers and by protecting the prized cheap labor pool in the Bemba-speaking
northeast from foreign competition.
All of these policies and actions, as we have seen, had important effects on shaping the
contours of the contemporary Zambian linguistic map. Yet, as I have just suggested, the shape of
that map must be seen not just as a product of colonial-era policies and actions but, even more
interestingly, as an externality of decisions made for entirely different purposes. That the shape
of this map would have such important implications of the character of contemporary Zambian
politics is one of the great ironies of the colonial legacy.
30
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
Bibliography
Berger, Elena L., 1974, Labour, Race and Colonial Rule: The Copperbelt from 1924 to
Independence (Oxford: Clarendon Press).
Bettison, David G., 1959, “Numerical Data on African Dwellers in Lusaka, Northern Rhodesia,”
Rhodes-Livingstone Communication, no. 16.
Colson, Elizabeth, 1962, The Plateau Tonga of Northern Rhodesia (Manchester: Manchester
University Press).
Doke, C. M., 1945, Bantu: Modern Grammatical, Phonetical and Lexicographical Studies since
1860 (London: International African Institute).
1961, “Scripture Translation into Bantu Languages,” in C. M. Doke and D. T. Cole,
Contributions to the History of Bantu Linguistics (Johannesburg: Witwatersrand University
Press), pp. 108-114.
Davis, J. Merle, 1967, Modern Industry and the African: An Enquiry into the Effect of the Copper
Mines of Central Africa upon Native Society and the Work of the Christian Missions, 2nd
edition (London: Frank Cass). [first edition published in 1933]
Franklin, Harry, 1950, “The Saucepan Special” (Lusaka: Government Printer).
Gadsden, Fay, 1992, “Education and Society in Colonial Zambia,” in Samuel N. Chipungu,
Guardians in their Time: Experiences of Zambians under Colonial Rule, 1890-1964 (London:
Macmillan), pp. 97-125.
Gann, L.H., 1958, The Birth of a Plural Society: The Development of Northern Rhodesia Under
the British South Africa Company, 1894-1914 (Manchester: Manchester University Press).
Gourevitch, Philip, 1998, We Wish to Inform You that Tomorrow We Will be Killed With Our
Families: Stories from Rwanda (New York: Farrar Straus and Giroux).
Greig, Jack C. E., 1985, Education in Northern Rhodesia and Nyasaland: Pre-Independence
Period, Oxford Development Records Project Report 13, (Oxford: Rhodes House).
Hailey, William Malcolm Baron, 1950, Native Administration in the British African Territories
Part II “Central Africa” and Part IV “A General Survey of the System of Native
Administration” (London: HM Stationery Office).
Hall, Richard, 1976, Zambia 1890-1964: The Colonial Period (London: Longman).
Harries-Jones, Peter, 1965, “The Tribes in the Towns,” in W.V. Brelsford, The Tribes of Zambia
(Lusaka: The Government Printer), pp. 124-146.
Hay, Hope, 1947, Northern Rhodesia Learns to Read (London: The Livingstone Press).
31
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
Henkel, Reihnard, 1985, “Mission Stations in Zambia: Their Location and Diffusion Patterns,”
Zambian Geographical Journal 35, pp. 1-18.
Hellen, John A., 1968, Rural Economic Development in Zambia, 1890-1964 (Munich: Weltforum
Verlag).
Johnston, Harry H., 1897, British Central Africa: An Attempt to Give Some Account of a Portion
of the Territories under British Influence North of the Zambezi (New York: Edward Arnold).
1919, A Comparative Study of the Bantu and Semi-Bantu Languages (Oxford: Clarendon Press).
Kasfir, Nelson, 1976, The Shrinking Political Arena: Participation and Ethnicity in African
Politics, with a Case Study of Uganda (Berkeley: University of California Press).
Laitin, David D., 1992, Language Repertoires and State Construction in Africa (Cambridge:
Cambridge University Press).
Luchembe, Chipasha, 1992, “Ethnic Stereotypes, Violence and Labour in Early Colonial Zambia,
1890-1924,” in Samuel N. Chipungu, Guardians in their Time: Experiences of Zambians
under Colonial Rule, 1890-1964 (London: Macmillan), pp. 30-49.
McCulloch, Merran, 1956, “A Social Survey of the African Population of Livingstone,” RhodesLivingstone Paper, no. 26.
Meebelo, Henry S., 1971, Reaction to Colonialism: A Prelude to the Politics of Independence in
Northern Zambia, 1893-1939 (Manchester: Manchester University Press).
Moore, Henrietta L. and Megan Vaughan, 1994, Cutting Down Trees: Gender, Nutrition and
Agricultural Change in the Northern Province of Zambia, 1890-1990 (Portsmouth, NH:
Heinemann).
Mytton, Graham, 1978, “Language and the Media in Zambia,” in Sirarpi Ohannessian and
Mubanga K. Kashoki, Language in Zambia (London: International African Institute).
Ohannessian, Sirarpi and Mubanga K. Kashoki, 1978, Language in Zambia (London:
International African Institute).
Perrings, Charles, 1979, Black Mineworkers in Central Africa: Industrial Strategies and the
Evolution of an African Proletariat in the Copperbelt, 1911-1941 (New York: Africana
Publishing Co).
Philpott, R., 1945, “The Mulobezi-Mongu Labour Route,” Human Problems in British Central
Africa, 3, pp. 50-54.
Posner, Daniel N., 1998, The Institutional Origins of Ethnic Politics in Zambia, Ph.D.
Dissertation, Department of Government, Harvard University.
Prins, Gwyn, 1980, The Hidden Hippopotamus: Reappraisal in African History: The Early
Colonial Experience in Western Zambia (Cambridge: Cambridge University Press).
32
Origins of Linguistic Divisions in Zambia
Daniel N. Posner
Ragsdale, John P., 1986, Protestant Mission Education in Zambia: 1880-1954 (Cranbury, NJ:
Associated University Presses).
Ranger, Terence, 1989, “Missionaries, Migrants and the Manyika,” in Leroy Vail, The Creation
of Tribalism in Southern Africa (Berkele y: University of California Press), pp. 118-50.
Richards, Audrey, 1939, Land, Labour and Diet: An Economic Study of the Bemba Tribe
(London: Oxford University Press).
Roberts, Andrew, 1976, A History of Zambia (New York: Africana Publishing Co.).
Rotberg, Robert, 1965, The Rise of Nationalism in Central Africa: The Making of Malawi and
Zambia, 1873-1964 (Cambridge: Harvard University Press).
Siegel, Brian, 1989, “The 'Wild' and 'Lazy' Lamba: Ethnic Stereotypes on the Central African
Copperbelt,” in Vail, The Creation of Tribalism in Southern Africa (Berkeley: University of
California Press), pp. 350-371.
Snelson, Peter, 1974, Educational Development in Northern Rhodesia 1883-1945 (Lusaka:
Kenneth Kaunda Foundation).
Spitulnik, Debra, 1992, “Radio Time Sharing and the Negotiation of Linguistic Pluralism in
Zambia,” Pragmatics 2, pp. 335-354.
Wolpe, Howard, 1974, Urban Politics in Nigeria: A Study of Port Harcourt (Berkeley: University
of California Press).
Young, Crawford, 1976, The Politics of Cultural Pluralism (Madison: University of Wisconsin
Press).
Zambia National Archives, various files.
33